Frequently asked questions

Questions people ask about Trackapt, answered the way they were answered. Each answer stands on its own; the page it belongs to is linked where there is more to read.

Mining: getting started

In what order do I actually use the Mining module?

Seven parts, each ending with something the next one needs.

1. Ground to stand on — a Tracker Group with Used by ticked for the technical work it carries, an item for the pit, and the mine grid named once.
2. Exploration — import collars, downhole surveys and interval tables, Assemble before you file, then Validate the whole register.
3. Interpretation — draw wireframes section by section and make them close.
4. Resource — define the block model, fit the variogram, estimate, and read the grade-tonnage curve.
5. What is worth mining — survey the topography, optimise a pit out of the model, and Make it a design. Underground, lay out stopes instead.
6. The mine as built — pit, water, stages, waste dumps, haul roads and ramps.
7. The plan — sources, processes and capacities, a horizon, compute, and publish each figure onto the parameter it becomes the plan for.

Then it loops: the shifts file against the published plan, the surveyor measures what moved, the plan is reconciled against the ground and the weighbridge, and the next period is planned from ground a month newer.

You do not have to start at part 1 — an operation already running usually starts at part 7 and fills the earlier parts in behind it. See From exploration to production for the whole road with every step.

Where do I start if I have never used it?

Part 5 of the walkthrough, backwards: import one survey of a dump you already know and check it against the ground you remember, then read cut and fill against last month's survey of the same item, then reconcile that against the weighbridge. Only then bring the drillholes in. Measuring something you can check by eye first is what makes the rest believable. The full fortnight is at the foot of From exploration to production.

What is the difference between Professional and Modular?

They are the two ways Mining can open for you, chosen under Mining view in the Appearance dialog (Themes in the header). Professional, the default, opens Mining as the workspace: the sidebar has one Mining entry, and a register or record opens in a window over the pit rather than on a page of its own. Modular lists every Mining screen under Mining in the sidebar and opens each register and record on its own screen, landing on Surveys; the workspace is still there, as the first entry. The choice is kept in your browser, for you and this account. See How Mining opens.

Why is Design blast greyed out in the workspace?

Designing a blast needs access level 4, by every way into it: Design blast and New blast on the Blast tab, Blast this block and Edit blast in the picture, and the plus on the Blasts register. Below level 4 the commands are greyed, and resting the pointer on one says Designing a blast needs access level 4.; the register's plus is not shown at all. Design blast is also greyed in a browser without WebGL, which its tooltip says, and with no working chosen, which it does not: choose a working in the Explorer first. See The workspace.

I can't see the blast holes under the ground in 3D. Where are they?

Under the survey. A survey at 100% opacity is solid, so whatever lies below the ground — holes, solids, a design you are drawing — is hidden under it. Select the survey, press Colour in Properties (or right-click its row for Colour and opacity…) and lower its Opacity. While you design a blast, Show holes, Show tie-up and Show outline hide parts of the design itself. See Seeing below the ground in 3D.

Mining: planning

What does "To the best order" do?

It is the third way a plan can be scheduled, and the one that runs the optimiser. The other two walk the pit one period at a time, top bench first: "To a product target" takes benches until the period's target is reached, "To capacity" takes benches until a process runs out. Both decide period one fully before looking at period two.

"To the best order" plans every period together. It works out what fraction of each bench to mine in each period, holding to every capacity, to the rule that nothing is dug from under rock that is still there, and to any target you insist each period reaches. Then it picks the order that gets the most out of whatever you asked it to maximise: product, value at your prices and costs, or net present value, which counts product taken sooner for more.

The result may leave a bench half-mined for a month, or strip more waste early so later months yield more ore, where the simpler walks never would. See Optimising the schedule.

What is a period?

One scheduling step of the plan. The cadence, the start date and the number of periods on the plan card define them: monthly, starting 1 October, three periods is October, November and December. Every period has its own capacities, since a fleet resolves to more tonnes in a 31-day month than a 30-day one, and its own result. Publishing writes one plan figure per period onto each parameter, and Trackapt spreads that figure over the readings that fall inside the period. The plan does not decide which day within a period a bench is taken; that is the general planning module's job, not the mine plan's.

What is the horizon?

The whole span the plan covers: the periods taken together, from the first day of the first to the last day of the last. In the two simpler modes it barely matters, since each period is scheduled on its own. In "To the best order" it is the thing being optimised: the objective is scored over the whole horizon at once, which is what lets the optimiser trade one period against another. A one-period horizon leaves it nothing to reorder. The horizon is also the limit of what the plan knows: benches left at its end are not weighed at all.

Two months into a twelve-month plan things did not go to plan. Can it be adjusted?

Yes, by re-planning, not by editing. The optimiser has no memory of its earlier answer. It always solves fresh from the ground as it stood on the plan's as-of date, so the thing that absorbs what actually happened is the next survey pickup: benches the pit has taken are gone from the reserve, whatever the plan expected.

Add the pickup to the running surface, then press Re-plan on the published plan and pick the period to start from. A new plan opens as a draft over the rest of the horizon, with everything copied. Set its as-of date to the pickup, adjust what changed, compute and publish. The old plan keeps its first two months exactly as published, as history, loses the rest, and is marked closed. See Re-planning part-way through and, for the chain of plans that builds up, The budget against the latest plan.

How does an approver know a plan is waiting on them?

They are told. When a plan reaches a desk, every approver at that stage gets a notification on their subscribed devices — the same way a form awaiting their signature tells them — and when it is sent back or published, the planner who asked is told, with the reason where there is one. On the Plans register a plan waiting on you reads Awaiting your approval, the header counts how many are, and Show → Awaiting my approval narrows the register to those. See Approval before publishing.

Can I approve a plan with no signal?

Yes. Approve and Decline queue on the device like a form filed underground and go up the moment it next has a connection. The plan on screen stays as it was until then, because nothing has changed on the server yet. See Working offline.

I approved a plan and was told someone else had just changed it. Did my approval count?

Not yet — nothing of yours was written. Two approvers at one desk signing within the same second both sign; you only see Someone else changed this plan a moment ago. Reload it, then try again. when the plan moved under you in a way that changes what you were approving: it was declined, withdrawn, or published by the last signature while yours was on its way. Reload the plan. If it is still waiting on you, approve it again; if it has moved on, there is nothing left for you to sign. Tapping Approve twice signs once: the second tap is told you have already approved — or, where your first tap finished that desk, that it was recorded and the plan has moved to the next desk. One tap never signs two desks, even where you are an approver at both: open the plan again to approve at the next one.

You see the same sentence where the planner withdrew the request, changed the plan and asked again while you were still reading the earlier one. Your Approve or Decline is held to the request that was on your screen when you pressed it — also when it was queued with no signal and sent hours later — so it is refused rather than signing, or on the last desk publishing, figures you never saw. Reload the plan, read what it says now, and decide again. This needs the page to be up to date: one left open since before an update of Trackapt is only protected once it has been reloaded. What it does not tell you is that a colleague signed while you were reading — that changes nobody's figures, and your approval simply joins theirs. See Approval before publishing.

A plan says "publish cut short". What happened, and what do I do?

Publishing marks the plan published and then writes a figure for every period onto the parameters you named. Publish cut short means the writing stopped part-way — the server restarted, or its connection to the database dropped — so the plan reads as published while some of its figures never arrived. The mine reads a missing figure as zero tonnes planned, which is why the plan's page says so in red and the Plans register shows Publish cut short in the Status column instead of Published (type publish cut short into the table's filter to find every such plan).

Unpublish the plan, then publish it again. Unpublishing takes back whatever figures did get written; publishing again writes them all. Until then Re-plan is not offered, because a re-plan is cut from the published periods and these are incomplete. Trying to publish over it without unpublishing first is refused with the same instruction.

For the few seconds a sound publish takes, the plan reads Publishing and its page says the figures are being written; the page checks again by itself and settles on Published. See Publishing.

Can the optimiser handle several pits in different stages, and decide which to attack?

Yes, on a plan to the best order. Add the other pits under the first pit's sources, each with its own ground, model and shape. Mark the plant and any shared fleet as processes that serve every pit, and each pit's own excavator as a process that serves that pit only. The optimiser then decides which pit to attack, for how much and for how long, to get the most out of the shared capacity. Nothing forces one pit to wait for another; the shared capacities are what tie them.

The result shows the split by pit, and each figure can be published as the plan total or as a pit's own share, onto that pit's own parameters. The two simpler modes take one pit only: walking top bench first through one list, they would finish the first pit before touching the next. See Several pits, one plant.

What is "greedy"?

The name for how the two simpler modes decide. A greedy method takes the best-looking step right now and never looks back or ahead: top bench first, one period at a time, every capacity refilled at the period boundary. It is fast, easy to check by hand, and usually right for a single pit, where the rock above must come off before the rock below anyway.

It cannot lend capacity forward, so spare haulage in a drilling-limited month is lost. With two pits it would finish the first before touching the second. "To the best order" is what fixes that. The greedy schedule still does three jobs inside the optimiser: it is the starting point the solver improves from, the floor the solver's answer must beat, and the fallback when no feasible schedule exists, in which case the result says so.

What is a band, and what is a lean bench?

A plant runs best on ore of a fairly steady grade, so a mine gives the planner a range its feed must average inside each period. That is the feed grade band on the Objective card: "at least" and "at most", either of which may be left blank, measured as recovered product per tonne of ore so two pits estimated in different grade units compare on what a tonne actually yields.

A lean bench is one whose ore falls below the band's floor on its own; a rich one sits above it. Neither is waste. A lean bench simply cannot be fed alone without dragging the period's average under the floor, so it has to be mixed with rich ore. The optimiser blends fractions of several benches to keep every period inside the band. The greedy walk does it more crudely, taking of a lean bench only as much as the surplus the rich benches built can absorb. See Blending, opening a pit, and the longer haul.

Will the waste fit in the dumps?

Add the dumps to the plan, in the order they are tipped, each with its dump design and what is already in it, and give the pit a swell factor. Compute then turns every period's waste tonnes into loose cubic metres — a tonne of rock at 2.7 t/m³ and a swell of 1.3 is 0.48 loose m³ — and tips them into the dumps in order. The result shows what each dump took, what is left, and the period it fills. A plan to capacity stops when the dumps are full and names the dumps' space as what stopped it; a plan to a target says, in red, how much waste has nowhere to go; a plan to the best order will not plan more waste than the dumps hold. See Waste dumps, and whether the waste fits.

How is a dump's capacity worked out?

From its design: the footprint, the lift height, the face angle (the angle of repose) and the berm, built lift by lift upward and inward until the top closes. The capacity is the space between that designed top and the ground the design names, inside the footprint, measured exactly against the survey — so on a hillside the uphill side of a lift holds nothing, which a spreadsheet frustum would count. It is in placed cubic metres. See Waste dumps.

Do I have to type a dump's footprint corner by corner?

No. If the area has already been set out in Survey as a boundary — which is usually how a dump allocation is agreed — pick it under Take a surveyed boundary on the design and its corners become the footprint. Boundaries are stored on the mine grid, the same frame a design is stored in, so nothing is converted and nothing is retyped. The same control fills a pit floor. It is not offered for a road's centreline, because a boundary carries no levels and a centreline is nothing without them. Drawing on the plan and pasting coordinates both still work. See Waste dumps.

Why does my dump come to a point at the top?

Because nothing told it not to. A dump is designed upward and inward, and stepping in lift after lift closes the outline — which is fine as arithmetic and useless as a plan, since a truck has to reverse to the edge, tip and turn round. Give the dump a least top width, about 40 m for a rear-dump fleet, and the design stands at the last lift whose crest is still that wide and reports Top as the reason it stopped. A dump with none set is built until its outline closes, and says in its message that the top is too small to work. See A dump has to be worked.

How do the trucks get to the top of a dump?

Up a ramp the design draws. Give the dump a ramp width and a ramp grade and it spirals a haul road up the outside of the faces, climbing each lift at that grade and crossing each berm still climbing, so it reaches every lift's toe exactly at that lift's level. Each crest it passes is a landing — the level tipped at while that lift is built. The report gives the length, the grade achieved, the laps round the dump and the levels it lands on; the 3D view draws it; and it leaves with the design on its own layers and as a body.

The dump ramp says it would cross itself. What does that mean?

That the dump is not far enough round for that climb. A lift's rise divided by the grade is the running it needs — a 10 m lift at 8% is 125 m — and if that is more than the dump measures round at that lift, one lap is not enough and the ramp would run over itself on the same face. Flatten the grade, lower the lift, or widen the footprint. Laps across different lifts are ordinary and are simply reported.

Does a dump's tipping pad change what a plan thinks it holds?

Yes. The plan reads the same least top width the design does, so a dump that stands two lifts short because nothing could tip on the third holds two lifts' worth in the waste balance too. A dump with no least top width set is planned at its full shape.

What does the plan say about rain?

Switch Plan for water on, type the site's mean rainfall by month and the storm the pit is designed for, and say what catches the rain in each pit and what pumps it out. Every period then shows its runoff against what the pumps can clear, and the design storm against the sump and the floor of the lowest bench reached — with the depth it would flood. A period whose runoff outruns the pumps is marked in red and the warning says by how much. Rain never stops a plan; it names the month so the low benches can wait for a drier one, or the pumping be added. See Water: rain and groundwater.

Does the plan allow for groundwater and the water table?

Yes, on the same dewatering block as the rain. Give the pit a steady inflow in cubic metres an hour — seepage and springs, as the hydrogeology report puts them — and, where the pit goes below the water table, the water table as a level with a rate per metre below it. Every period reads how far the floor it has reached sits below the table and takes that many metres' worth, all day, rain or no rain; a floor above the table takes none of it. The result shows the groundwater beside the runoff, the pumps are judged against the two together, and the storm carries the groundwater that keeps coming while it falls. Groundwater with no pumps to move it is marked in red like a shortfall, because that is what it is. See Water: rain and groundwater.

Does the mine design include water control?

The pit design carries a sump — its depth and plan area, dug below the floor at its centre — and a diversion drain run round the pit a stated distance beyond the top crest. Both are drawn in the 3D view and reported beside the benches: the sump's capacity, the level of its bottom, and the strip inside the drain that still drains into the pit. A plan for the pit reads the sump and the strip from the design where the planner leaves them blank, so the pit is dewatered as designed. The water table itself, and how much the pit takes from it, is typed on the plan, not the design. See Water in the pit.

Mining: geology

What is a hull?

The smallest shape that contains a set of points with no dents in it. Hammer a pin into a board for each point, stretch a rubber band right round the outside, and let go: the band snaps onto the outermost pins and pulls straight between them. That taut outline is the convex hull.

Convex is the no-dents part — pick any two spots inside the shape and the straight line between them stays inside. A crescent, a star or a kidney shape are concave, so none of them can be a hull.

On a wireframe section the intercepts above the cut-off are the pins. A hull round them is fast and will never leave ore out — but it cannot bend inward, so a crescent-shaped lens comes back as the block that fills the crescent, with the barren middle counted as ore, and two separate shoots on one section get one band round both. That is why a body built from the drilling is a first pass: a geologist's outline curves in where the ore stops, and a rubber band cannot. See A first pass from the drilling.

Are wireframes drawn by hand, or built from the drillholes?

Both, and in that order. On an item that carries drilling, the wireframe editor offers Build from drillholes: give it the interval table, the field, a cut-off, the section bearing (normally the strike), the section spacing (usually the drill spacing) and a minimum half-width, and Build the sections reads every hole, takes the intervals at or above the cut-off, cuts them into sections along that bearing and puts a convex hull round the intercepts on each one. What comes back replaces the section boxes, and you edit it from there.

It is a first pass, not an interpretation, and it says so every time it runs. Interpretation is the one step in the chain that is a judgement rather than a calculation — a geologist draws an orebody using structure, logging and everything they know about the deposit, none of which is in the assay table. A hull per section cannot be concave (a crescent lens comes back as the block containing it, barren middle counted as ore), joins two shoots on one section into one body, lofts straight through a fault offset, and stops at the outermost hole. All four make the solid bigger in the middle and smaller at the ends, which is the wrong direction for a resource.

So take the first pass to save the afternoon of pasting coordinates, then edit the sections into the shape the geology actually is. See A first pass from the drilling.

Can I optimise a pit with no signal?

No. Unlike filing a form or approving a plan, the pit optimisation runs on the server and needs a connection — it is the one screen in the module that will not work out of signal. Everything either side of it is offline as usual: the model is already on the device once loaded, and the design the answer becomes is filed like any other record. See Working offline.

"Optimise a pit" says my model is too big. Why, and what do I do?

Past a million blocks the run is refused rather than part-solved. The reason is that a pit optimised over part of a deposit looks exactly like a pit optimised over all of it — same table, same outline, same confident stripping ratio — so there is no safe way to quietly do less of it.

Re-estimate the model on a coarser lattice and run it again. Bear in mind the block size is also what decides the flattest wall the optimiser can express, so going coarser trades wall detail for reach; the dialog says what slope the pit actually stands at once it has run.

Mining: working with other packages

What shape does the file I am uploading have to be?

Whatever the Template button beside the file input hands you. Every file input in Mining has one, and it gives you the file that reader takes — the column names it looks for, with two or three rows filled in as an example. Take it, fill it in, or send it to whoever is producing the data.

It is worth doing before the first import rather than after the first failure: the readers are forgiving about the order the columns come in and strict about what has to be there, so a file that is nearly right loads without complaint and matches nothing. The survey blank follows the instrument profile you picked, because the profiles differ in which column the point name is in and which of the two coordinates is the northing. See A blank to start from.

My rover's pickup reads as nought ground points

Check the Columns read as line in the import dialog against one row of the file. The usual cause is a rover that numbers every record before naming the point — a Hi-Target writes 1,B092009,7835077.665,433982.989,712.293,set_base — so the coordinates sit one column further along than a total station's pickup puts them. Work it out from the file finds that by itself; if you have picked a layout by hand, pick Number, Name, Northing, Easting, Level, Code instead.

The message tells you which of the two it is. No rows read as points means the layout does not match the file at all — nothing was read, so nothing was rejected on its merits. Too few ground points means the file was read and what came out was control, feature lines, or shots too far apart to be one piece of ground; the summary below it says which.

The import says a shot is metres above the ground — is my survey wrong?

Probably that one shot is. Every ground shot is compared with the ground immediately around it, and one standing metres clear of its neighbours is the shape of a rod left long or a level typed with two digits the wrong way round. It is set aside and listed; tick Include them anyway if you know it is real.

A shot metres below its neighbours is a different matter. A sump, a drain, a void or an over-dug toe is a hollow that nothing around it agrees with, which is exactly what the check keys on, and all four are ground you were right to pick up. Look at those on the plan before leaving them out.

Nothing was flagged on height — does that mean my levels are good?

No, and the dialog says so in metres. The check works against the ground's own roughness: on a graded bench it can see an error of a metre or so, on broken ground only a much larger one, and an error of a fifth of a metre is invisible to it anywhere. It finds blunders, not precision. A piece of ground carrying fewer than twelve shots is not judged at all, and the dialog says that too.

My control is numbered CP1, CP2, CP3 — do I have to name every one?

No. The reader takes a numbered code as the code it is numbering, so CP7, TBM3 and BASE1 are read as control on their own. It only strips the number where what is left is a code it already knows, so an invented code that happens to end in a digit stays ground.

If you have told the reader what a code means, that answer covers its numbered forms as well — but only towards ground or a breakline. Say CP is ground here and CP1..CP8 follow it. Say CP is control and only the exact code CP moves: ground taken for control vanishes out of the surface with nothing to show for it, while control left in the ground stands up as a cone that gets queried.

The one place none of this applies is the layout whose fifth column holds the name of a feature rather than a code. BM1 and BM2 there are berm 1 and berm 2 far more often than they are benchmarks, and a name read wrongly costs the whole feature line.

The import is asking whether BM is a control point or a feature. Why can it not tell?

Because on that file it genuinely cannot. The layout is the one whose fifth column holds the name of the feature rather than a code, so a crest the surveyor called BM looks exactly like a benchmark called BM, and points sharing a name are read as one line. Nothing about the file settles which it is, so you are asked rather than guessed at. Answer once and it is remembered for this site.

Can I skip that question and file the survey anyway?

No. File survey stays greyed until every such name has been answered, and the footer names the ones outstanding. There is no way past it.

The reason is a measurement. A site's control network is routinely booked as a dozen scattered marks under one string name, and read as ground those become one line threaded across the whole property. On a 64-shot floor with a ring of twelve marks 25 m clear of it, the fill against a datum came out 159% over the honest figure; at 60 m it was 648% over. Nothing else in the import catches it — the separate-pieces-of-ground check cannot see it on this layout, and the longest-triangle clip only helps when the network is far enough out to be obvious. A crest read as control is quieter still: simply absent from the surface. Both mistakes are silent, and a silent wrong volume is worth more than one click to avoid.

Why does it block every name, even one that is obviously a crest?

Because deciding which names are obvious would be the same guess the question exists to avoid. If the software could tell a crest called BM from a benchmark called BM, it would not need to ask at all. One rule that always applies is also a rule you can hold in your head; "some of them have to be answered" is not.

Why was I not asked about the benchmark I shot five times?

Because nothing was at stake. A run of shots that never moves is repeat observations of one point, whatever it is called, so it is held as control without asking — even where a crest in the same file shares its name. The same goes for a name carried by a single shot: there is no line to lose. The question is raised only where getting it wrong would take a whole feature line out of the surface.

Which card menu holds exports and imports?

Card menus are named for what you are trying to do: Export for anything that writes a file you keep or sends the figures on, Import for anything that reads one in — including the blank template — Edit for changing what is on screen, and View for looking at it differently. The same names mean the same things on every screen. See Card actions.

Does Trackapt work with Deswik, or Deswik NOVA?

Yes, in both directions, by file and by API. NOVA schedules a reserve it is given; Trackapt reads the reserve off its own survey, block model and design, and hands it over as Reserve as CSV, or hands NOVA the pit as bodies, the block model as every block with its definition, and the topography as triangles so NOVA can cut its own. NOVA's schedule comes back as CSV through Import a schedule, or through the exchange API, and the plan becomes an imported plan whose every figure is read off NOVA's takes — then published here, where the dashboards judge it. Trackapt is not a scheduler of NOVA's depth, and NOVA is not a survey, reconciliation and reporting system; each does its half. See A plan, to and from a scheduling package.

How do I get a schedule from another package into a plan?

Compute the plan once so it has a reserve, send the reserve away, and when the schedule comes back as CSV use Import a schedule under Import. Rows are matched to the reserve by pit, bench and stage — or by the reserve's own unit column carried through — never by position, and either a fraction or tonnes per period is read. The dialog says how many rows matched and which did not. Taking it makes the plan an imported plan and computes it at once. See A schedule decided elsewhere.

What is an imported plan?

A plan whose schedule was decided elsewhere: its takes are exactly what the file said, period by period, and Compute reports what those takes cost — tonnes, grade, product, every process's demand against its capacity, truck hours, feed qualities, the dump balance — without changing a fraction. A capacity it exceeds is red, a bench taken after the one below it is reported, and a row that asked for more than was left takes what was left and says so.

What is the exchange API, and how do I get a token?

A way for a program — a scheduler, an ERP, a script — to pull a plan's reserve, schedule and demands, a block model, a survey or a design, and to push a schedule back, without anyone downloading files. The account holder or a Level 6 user opens Integrations on the plans register, makes a token named for the package, and hands over the secret, which is shown once. The token goes in Authorization: Bearer tk_…; read reads, write files a schedule. Revoke it there when the integration ends. See The exchange API.

Can another system be told when a plan is published?

Yes: register a webhook under Integrations, and every publish or unpublish is posted to it as JSON — the event, when, and the plan's id, name and status — signed with an HMAC under the hook's own secret in X-Trackapt-Signature. Verify the signature over the raw body before trusting it. A hook can be paused, and is removed after asking.

Why does the block model CSV carry ix, iy and iz?

Because a scheduler rebuilds the lattice from them. A centroid tells a receiver where a block is; the indices tell it which block it is, so a model that comes back — re-blocked, re-estimated, cut against solids — lands on the same cells. Columns are found by heading in every reader, so a package that ignores them is unaffected. Every block as CSV writes the blanks too, and Definition as JSON the lattice alone.

Can a plan use a pit that was drawn in another package?

Yes. On the design's editor, under As drawn elsewhere, read in the pit's toe and crest rings as a Surpac string file or a DXF; the pit is then those benches, and the slope rules only describe it. A plan for that pit reads exactly the benches that were drawn. Back to the rules drops them, after asking. See A pit as drawn elsewhere.

What is a scenario?

A draft copy of a plan kept beside it, made with New scenario, so one thing can be changed and the two compared: Compare scenarios lays the original and every scenario side by side, over the horizon and period by period. A scenario is a plan in every other respect; the one that wins is the one that is published. See Scenarios.

How is a capacity from a truck cycle worked out?

From the trucks, the road and the shifts, the way a haulage engineer does it. Each truck class has a payload and its speed against total resistance — grade plus rolling resistance — loaded and empty; the route is segments of length and grade, read off a filed haul road if there is one; add the loading and dumping time and the cycle follows, and the tonnes the fleet can carry round it in the period is the capacity. The result reports the trips, the truck hours and, where the class states litres an hour, the diesel and the CO₂. See Haulage from the road.

Can a declared capacity change from one period to the next?

Yes. Tick Varies by period on a declared capacity and type a figure per period, or paste a column from a spreadsheet; a period left blank takes the single figure. Three trucks in March and five from April is one plan, not two.

Can the plan blend on more than the grade?

Yes. Under Sources, Qualities names up to eight more attributes of the block model — sulphur, silica, moisture — each carried into the reserve as a tonnes-weighted mean. Each can carry a feed band; to the best order the bands bind and the optimiser blends pits, piles and periods to hold them, while on the walks and an imported schedule a breach is reported. A recovery curve against grade replaces the single recovery, and Several piles gives up to six stockpiles, each with its own grade band, so high-grade and marginal rock are piled apart. See Qualities, blending and several piles.

Mining: underground

Does Trackapt do underground mining?

Yes, for design and planning. Headings are designed in Mine designs — a decline, a drive, a crosscut or an ore drive along a centreline, with the profile the machine cuts — and a plan whose shape is Stopes and headings lays the orebody out into stopes, reads the development that reaches them, and schedules the two together against the same capacities a pit plan uses. What it does not do is stope shapes wrapped to a rolling orebody, ring design inside a stope beyond the existing drill rings, or ventilation. See Underground: stopes and headings.

Can I see a stope layout before committing a plan to it?

Yes. On Block models, with a model open, Lay out stopes runs the layout on its own and shows what it comes to: how many stopes pay for themselves, the tonnes fed and how much of that is wall slough, the grade the mill sees, the worth before development, and a row per stope. Run it at several prices to see whether the layout holds or whether it is one bad quarter from being a different mine. It files nothing, because a layout is what a frame and a set of costs imply and both live on the plan. See Laying stopes out of the model.

Why was my stope frame laid out at different figures from the ones I typed?

Because a stope is built from whole model blocks. A 25 m sublevel on a model of 10 m blocks is laid out at 30 m, and a 3 m minimum width at 10 m. Both are reported where they happen rather than silently honoured. Where the frame matters more than the block size, the remedy is a model with smaller blocks.

How does it decide where the stopes go?

By what pays. The model is cut into a frame — a sublevel high, a panel long — and in each panel the plan takes the run of rock across strike worth the most, between the narrowest the equipment can mine and the widest the ground will stand. Every admissible width is examined, so no better one exists within the frame. A panel whose best run is not worth mining is left standing. The frame itself is not floated along strike, which is stated rather than approximated.

Why did my stope grade come out below the block grades?

Two reasons. Everything mucked from a stope is fed to the mill, including the waste caught between the ore — nothing is sorted underground — so a stope forced wider than the ore carries that waste at its own grade. And the walls slough: ELOS adds barren rock from both walls to every stope, whatever width it was taken at. A layout that ignored either would overstate every grade it reports.

What is ELOS?

Equivalent linear overbreak: how far a stope wall sloughs in, expressed in metres of wall. Half a metre on each wall of a 20 m panel over a 25 m sublevel adds 500 m³ of waste to that stope. It is the honest way to carry unplanned dilution, because it does not depend on how wide the stope itself is.

Can a plan mine a stope before the drive that reaches it?

No. By the end of any period, no more of a stope has been taken than of the development it follows. What the schedule does allow is progress in step — two thirds of a decline permits two thirds of the stope below it — because it is solved in fractions. To make access all-or-nothing, pin the heading to the period it is driven in, and every stope behind it waits for that period to finish.

Why will an underground plan not run to a target or to capacity?

Because those walks take benches top-down and underground has no top-down. A stope waits on the heading that reaches it, not on the ground above it, so the plan is made to the best order or read in as an imported schedule. The plan refuses the other two rather than producing a sequence that mines a level nobody has reached.

Why does the plan drill my decline at a different rate from the pit?

Because the decline has a development round of its own. Where a round is designed for a heading in Underground, the plan charges the heading's drill and explosives processes at that round's drilled metres and kilograms per tonne — a 5 × 5 m face with relief holes drills far more per tonne than a bench does. The plan takes the round approved last (or, with none approved, the round filed last) and lists it on the Drill and blast card beside the heading. It also names any heading with no round, which keeps the typed rates.

Where do I design a stope's rings?

From the mine plan, where the stope is laid out. Compute a saved plan of stopes and headings, then press Design rings beside the stope under Planned to fire in the selected period. The ring designer opens on the stope as the plan laid it out. Once saved, the plan charges the stope all the drilled metres and kilograms of its rings still to fire, each period the share of the stope it mines. Rings already fired are spent, and are not charged again. See Designing a stope's rings.

Is an ore drive ore or development?

Both, and the plan treats it as both. Its metres are development and its rock is hauled, but where you tell the plan what grade it is driven in, those tonnes are fed. On many mines it is the first ore the mill ever sees, months before a stope is ready.

Can Trackapt design a development round?

Yes. On Underground, state how the mine drills its development, record the rock domains, then Design a development round: choose the heading, where along it the face stands, the rock and the explosive, and the round is worked out and drawn on the face — Holmberg's four-section burn cut. See Designing a development round.

Why will it not design anything until I say how the mine drills?

Because a jumbo round and a handheld round are different rounds, and guessing which one a mine uses would design the wrong one with confidence. Both questions on Underground start unanswered on purpose.

Where do the rock figures come from?

From the mine's own rock domains. Each figure is marked measured, from the block model, typical, calibrated or stated, and every round shows those marks beside the figures it used. A round that needs a figure the domain does not hold says which, rather than assuming one.

Can I change the hole depth or the number of relief holes?

Yes. Type your own and the round is redesigned around it. What you typed is filed beside the figure the design worked out, so both stay on record.

How are a round's explosives taken off the magazine when it is fired every shift?

Each Log a firing files that shift's explosives: the design's, scaled by the holes actually drilled. The log's Magazine column says whether each firing came off the magazine — Filed, Partly filed, Not filed — and what was not filed is listed under the log with the kilograms to add by hand. A round that has been fired cannot be removed — Retire it when the heading moves on, and its firing log is kept.

Does Trackapt design a handheld stope face?

No. A handheld stope face is listed in the Blasts register, but Trackapt does not design it from the rock. Long-hole stope rings are designed: see Where do I design a stope's rings?

Mining: seams

Can Trackapt plan a thin chrome seam like the ones on the Great Dyke?

Yes. A seam of 25 to 30 cm that comes to surface along strike and dips gently into the ground is modelled in Seams from the holes that cross it — its floor, its true thickness and its grade, cell by cell — and planned with one of two shapes of its own: Seam strips from the outcrop, taken down dip until a strip stops paying for its waste, and Seam panels, mined underground between levels and reached by inclines. See A thin seam: panels and strips.

Why is a seam not a block model?

Because blocks cannot see it. A 0.3 m seam is 6 % of a 5 m block, so every block it crosses is mostly waste, and blocks thin enough to hold the seam are far more than a model may carry. A seam is a sheet, and everything that matters about it — how thick, how rich, how deep its floor lies — is in its own plane, so it is modelled as a grid in plan. See Modelling a seam.

How is the seam's thickness worked out from the holes?

Each hole's crossing is read between its two contacts, and the length down the hole is corrected to the thickness square to the seam using the seam's dip — which is fitted through the crossings themselves, or typed where there are too few holes to fit it. A vertical hole through a seam dipping 15° reads it 3.5 % long; an angled hole reads it longer still. See Modelling a seam.

Why is a seam's grade averaged by thickness?

Because a thin rich intersection and a thick lean one carry different amounts of rock. The grade is estimated through the accumulation — grade times thickness — and a cell's grade is its accumulation over its thickness, so 0.3 m at 40 % beside 0.6 m at 10 % averages to 20 %, not 25. See Modelling a seam.

What decides the grade the mill sees from a seam panel?

The stoping height. A panel is broken to at least the height people and machines can work in, and on a 0.3 m seam stoped at a metre, 0.7 m of wall rock goes to the mill with every square metre of seam. At 40 % in the seam the mill sees about 15.5 %. It is the figure a thin-seam mine lives or dies by, and the plan shows it rather than hiding it. See Seam panels.

How far down dip do the strips go?

Strip by strip, while each one pays for its own overburden and for the extra final wall it adds, and — where you set one — while its waste per tonne fed stays under the Most strip ratio. The first strip that fails ends the cut; the strips beyond it are listed, not taken. See Seam strips from the outcrop.

Can the overburden go back into the strips?

Yes: turn on Cast the overburden into the strip before. Each strip's waste then goes into the void of the strip beside it and asks nothing of the dumps. The first strip of each cut has no void beside it and still goes to the dump. See Seam strips from the outcrop.

Why will a seam panel plan not run to a target or to capacity?

For the same reason stopes will not: those walks take the reserve top-down, and panels underground wait on the incline that reaches them, not on the ground above them. Plan it to the best order, or read it in as imported. A strip plan does run to a target or to capacity — its top-down is down dip. See Seam panels.

Can one seam be mined from the outcrop and then underground?

Yes, as two plans on the same seam: a strip plan for the open cut, and a panel plan whose Crown pillar is set to the open cut's depth, so the panels start below it. Each plan reads the same seam; the strip layout's limit depth is what to set the crown pillar to. See A thin seam: panels and strips.

Mining: drill and blast

Can Trackapt design a blast for me?

Yes. Pick the bench, set the hole diameter, the product and the rock, and Draft the pattern lays the holes from the free face, drills each to the floor from the ground it stands on, and works out the explosives, powder factor, drill yield, fragmentation, vibration and flyrock. See Drill and blast.

How does the tie-up know which way is the free face?

It does not look for one itself. It takes the free face the pattern was laid from and starts the shot at an end of the row nearest it — row 1. That face is found on the surveyed ground: an edge of the block where the ground falls to half the bench within two burdens of the outline. If no edge does, the blast reads as a box cut and its rows run along the edge with the lowest ground beyond it. The Front row tile says which it found, and a box cut or a face standing out past the outline is said on the pattern card too. If it is wrong, pick the face's edges yourself under Front row, or add a firing point by hand in the tie-up. See The free face.

What is the back line, and why design from it?

The back line is the back of the blast: the crest of the berm or wall it stands in front of. Laying the pattern from it puts the back row a set distance in front of that line — the edge stand-off unless you type another — and every other row one burden toward the face, so the back of the dig is a clean line that respects the berm and gives a clearly defined pit. The front row takes what is left; the pattern card shows its burden to the face, and the front-row checks judge it. Holes are still named from the front: row A is the front row. Unless you mark the back line yourself under Front row, it is read off the outline — the edges facing away from the free face. See The back line.

Is the drafted pattern final?

No. It is a starting point. Move, add or remove holes, add filler holes where the blaster judges they are needed, give one hole its own stemming or product, or bring in the pattern the surveyor marked out. Every figure follows the holes you end up with.

Where did the set-up tiles go?

Once a pattern is drawn they fold away under The set-up, so the picture of the pattern comes first. Click the heading to open them again; the heading counts any tile that still needs attention, and your choice stands while you stay on the blast. The checks and the front row are never folded away. See Drill and blast.

How do I add a filler hole?

Two ways, and both end in the same form. Add a hole above the holes table: type where it is, leaving a figure blank to use the pattern's. Or Place a hole beside Draft the pattern: the pointer takes a crosshair over the plan and a click on the ground opens the form with the collar where you clicked. A click near a position the pattern laid that no hole stands at takes that position exactly, so a filler put back in a gap stands where the pattern wanted it; a click on open ground stands where it fell. Either way the hole is charged like every other one and shows in the table as Added by hand.

If it lands too near a hole already standing, the checks say so — see Why does the blast say two holes stand too close?.

How do I remove a blast hole from the pattern?

On the blast's Holes table, use the row's Remove control (the delete action beside Edit) and confirm. The hole comes off the pattern and every figure — explosives, powder factor, fragmentation — follows the holes left. The checks then measure the gap: where the ground the hole was to break is out of every other hole's reach, the checks card names the two row neighbours now carrying it, how far apart they stand, and the point to add a hole at, and Add a hole here on that warning opens Add a hole there. Save the blast to keep it; after approval the save also writes the removal in the change log, with the reason you type in Why the pattern changed.

A hole that has already been drilled is better kept: open it and mark it Abandoned instead, so it stays on the record but out of every total. Holes cannot be removed while the blast is awaiting approval or once it is fired. See Adjusting the pattern.

Why does the blast say two holes stand too close?

Because two holes closer than half the smaller of burden and spacing share one burden between two full charges: the powder factor in that ground is twice what the design chose, which over-breaks behind the face, throws off the collar, and can cut off one hole with the other. The checks measure it at the collar and at the toe — holes that stand clear at the surface can converge below — and name the pair, with Open … to move the hole, lean it off its neighbour or take it out. Both holes carry the warning in the Holes table and in the tooltip on the plan, in amber: both still fire, so it is advice rather than a fault. A wall row and the holes added to reach an island keep their own spacing against the pattern, since both are drilled tight on purpose — but a hole you place by hand is judged against them like any other, because a filler half a metre from a trim hole puts a full production charge inside the buffer zone against the final wall. See Adjusting the pattern.

What do the colours on the blast plan mean?

The holes are sandy: the darker dot is the collar, the line the hole projected to plan. A red hole has an error — it cannot fire as designed: nothing to charge, a collar past the final wall, a wet hole loaded with a product water spoils, a hard toe or no free face at the toe or too little rock in front of the charge, or the gap left beside it by a removed hole. An amber hole has only a warning — something to know while it still fires: an unsurveyed collar, a stand-off note, why a fix could bring it no further forward, or a hole standing too close to its neighbour. Rest the pointer on a hole to read what it is. Blue-green shading is the berm or island a card's eye points at; violet is the ground behind a berm's crest; green is the crest of the bench below. See the blast plan.

Why did the pattern leave holes off the crest of the bench below?

Because a block's outline can run past the top of this bench onto the top of the next bench down — along the foot of a wall, or where a bay of the pit wraps round. The ground there is a lip: at or a little under the crest you are blasting, the wall rising close behind it, and the ground in front falling away to below this blast's floor. That lip is the top of the bench below, and a hole drilled on it is on the wrong bench — it belongs to the shot that takes the bench below, working down from the top. So the pattern leaves a designed hole off it, counts it on its own card in What it takes with the RL range of the ground, and shades it green when you open the card's eye. An ordinary front row on a flat bench top is untouched — nothing rises behind it, or what does (a heap of muck, a windrow, a boulder pile) comes back down to the bench top rather than standing up as the bench above — and a hole you put there yourself is kept and warned, because it is yours to judge. See The crest of the bench below.

Can I bring in a drilling pattern from the survey?

Yes — file the surveyor's points as a pickup, then Pattern from survey on the blast. Each point inside the blast becomes a hole, drilled to the floor from its surveyed height. As drilled from survey instead matches the collars picked up after drilling to the holes they were drilled for.

If the same pickup also walked the crest or the floor, choose the pattern's code (such as PTTN) under Shots coded. Then only those shots become holes. See Holes from survey.

Can I change a blast after it was approved?

Yes, until it is fired. Each change after approval is written in the blast's change log — which holes, by whom, and why — so the pattern that was signed off and the pattern that was drilled are both on record. While it is Approved, a change to the pattern sends it back for approval where its group has an approval chain: type Why the pattern changed first, and the blast returns to Designed, its sign-offs dropped, to be sent again. With no chain it stays approved. Once it is Drilled or Charged, changes are logged only. See From approval to the muckpile.

Why did Trackapt say someone else changed this blast?

Because someone did, in the moment between your opening the blast and your pressing the button — a colleague saved an edit, signed, sent it back, or recorded the firing first. Nothing of yours was written. Reload the blast, look at what it says now, and do it again if it still applies. This is what stops two people's work from quietly overwriting each other: two approvers who sign together both sign, and a firing pressed twice is recorded and taken off the magazine once.

An approver also sees it where the blaster withdrew the request, changed the holes and asked again while the approver was still reading the earlier one. Approve and Send back are held to the request that was on your screen when you pressed, so neither lands on holes you never saw: reload the blast, read what it says now, and decide again. This needs the page to be up to date — one left open since before an update of Trackapt is only protected once it has been reloaded. A colleague's signature arriving while you read is not a change of this kind: the holes are the same, and your approval joins theirs.

An approver can meet two more sentences, both when a blast moves to the next desk of its chain while their approval is on its way:

• "Your approval was recorded at [the desk]. The blast has moved to the next desk — open it again to approve there." Your approval counted, and completed that desk. If you sit at the next desk too, open the blast and approve it there as its own decision: one press never signs two desks.
• "[The desk] was completed by another approver as you approved, so yours was not needed there…" A colleague's approval completed the desk a moment before yours arrived, so yours was not recorded. Nothing is wrong; open the blast again if the next desk is yours as well.

The same holds for underground blasts. See From approval to the muckpile.

Does firing a blast take the explosives off the magazine?

Where the product is linked to the magazine item under Explosives, yes: its kilograms are filed there as a reading on the firing day. A reading someone already typed for that day is never overwritten; the firing tells you what to add by hand.

A fired blast says "partly filed" or "not filed". What does that mean?

The firing is recorded first and the magazine is filed second, product by product, so a blast can be fired with only some of its explosives taken off the magazine. The blast's page says which under Magazine, and the Blasts register has a Magazine column you can filter on:

• Filed — every product came off the magazine. Nothing to do.
• Partly filed — some products came off and some did not.
• Not filed — nothing came off, and there are kilograms owed.
• Nothing to file — filing was switched off for this firing, or no product fired is linked to a magazine item. Nothing is owed.
• Filing cut short — the firing was recorded but the filing never finished (the server restarted mid-way), so the blast cannot say what reached the magazine.
• Filing unconfirmed — the filing was started moments ago and is not marked finished. A blast does not keep when its filing started, so one still running cannot be told from one cut short. Reload in a minute; if it then reads Filing cut short, treat it as that.
• Not recorded — the blast carries no record of how its filing went, and nothing shows as taken off the magazine. Not a failure, and worded so that filtering for Not filed leaves it out.

For partly filed and not filed the page lists, in amber, a sentence per product that was not filed — no magazine item, an item no longer under Explosives or renamed, a magazine that no longer keeps the figure the product is filed onto, no access to the magazine, or a reading for that day that was typed or corrected by hand — with the kilograms to add to the magazine by hand. Add them on the magazine item's own reading for the firing day. For filing cut short, open the magazine item first and look at the reading for that day: add only what is missing. An underground round says the same against each firing in its log, and lists what is owed under the log. See The magazine.

Does the mine plan use the blast designs?

It can. Switch on Design from the pattern on a plan's Drill and blast card and each bench's drill metres and explosives come from its own pattern — in every mode, the optimiser's included. A computed period's blasts can be drafted straight from the plan.

How accurate is the fragmentation prediction?

It is Kuz-Ram, the industry's first estimate. It is only as good as the rock factor: reconcile a few blasts with a measured mean fragment and Trackapt suggests the rock factor your ground actually behaves as.

Where are the blast's settings — the block, the rules, the timing?

In the six tiles along the top of the blast. Each opens its section as a window over the page, so the pattern keeps the width of the screen; what you change applies as you type, and the pattern behind the window moves with it. A tile with a warning stripe has something that needs you.

How is each hole's length worked out?

From the ground and the grade — you never type it. A hole runs from the surveyed ground at its collar down to the grade under its toe, plus the sub-drill, divided by the cosine of its lean. The grade is the floor RL, or a grade surface chosen under The block for a floor that is not flat. The holes table shows each hole's lean and toe RL. See How long each hole is.

The front row says "hard toe". What does it mean, and can Trackapt fix it?

The face leans back or kept a toe, so a front hole has much more rock in front of it at the bottom than the burden it was designed for — the toe will not break out. Fix the front row leans each such hole towards the face, then brings it forward only as far as its charge keeps enough rock in front, and names any toe that still needs a toe hole or a heavier bottom charge. Where a bund, a hump or the crest's edge stopped the collar, it says the hole could not come further forward because a rig cannot stand there. Drilled holes are never touched.

Where it says instead that the hole could not come further forward because the charge would not rise above the toe there, the top of the bench falls away towards the crest: brought forward, the collar comes down, and under the same stemming the top of the charge would come down to the toe, leaving no charge above it to break it. That toe needs a toe hole or a heavier bottom charge. The hole's note gives the collar's RL, the stemming and the level the toe is judged at.

It does this as the pattern is drafted, by default, so a fresh draft comes out already fixed. To have hard toes only flagged, turn Fix the front row as the pattern is drafted off under the Front row tile. See The front row.

I drafted a pattern. Why does Checks still name holes?

A draft answers every check a design can answer before you see it. It leans hard toes, holds back a lean that would crowd the hole beside it, stems holes deeper under thin rock, and times each hole to fire after the hole ahead of it. What the draft put right lists every change, hole by hole, each with its own Undo. What Checks still names is the ground's to answer, and each check says how: a toe no lean or move can reach needs a toe hole, a heavier bottom charge or the rock in front dug out; no free face at the toe means the muck in front must be cleared; thin rock too low for stemming needs a deck or a lighter load. See What a draft solves by itself.

How do I find a hole a check names?

Under each check are links: Plan, 3D and Pictorial open a window on the holes it names, ringed and framed close in, and Open beside a hole opens it in the hole editor. The hole editor's header has the same three pictures. See Going to the holes a check names.

Leaning the front holes leaves the row behind a bigger toe. Should every row be leaned?

Leaning row after row, tapering back to plumb, does take the extra toe away, but it leaves the blast's back wall slanting, and the next cut on that bench starts with a toe it made itself. So a draft lays its front row as a leaned front row with an easer row behind it: the leaned holes take the toe at the face, and between each two where either is leaned an easer stands half a spacing across and half a burden back, in a row of its own, leaned half as far. The rows behind keep their burden and stay plumb, and every row is a row of one angle, so the crew can tell the front row from the easer row on the bench and tie each up for what it is. Only the holes whose toe needs it lean; on easy ground no hole leans and no easer is added. An easer fires half way from the later of its two leaned holes to the row behind, rounded up to the whole millisecond (with shock tube, the nearest connector not less than that). A mine that wants each hole leaned on its own with no easer row can still choose One hole a spacing apart under the Front row tile. Choose a denser bottom charge product to load every front hole heavier at the bottom, under a longer stemming (1.3 and 1.0 burdens by default, after Langefors and Kihlström; both can be changed).

A heavier front throws further. The checks show Front row flyrock and the warnings say how far to clear the blast; they also ask you to check how far the easers stand from their neighbours against your explosive supplier's least distance between charged holes, and name any pair too close for its delay. See An easer row and a bottom charge.

Why does one hole say both "hard toe" and "flyrock"? Isn't its burden too big or too small?

Both can be true, because they are two readings taken at different heights and not always the same way. The toe is read just above the grade, the way the hole breaks. The cover over the charge is read all the way up the charge, the way it is thinnest — flyrock goes to the nearest open ground. Each finding says which way it was read. A face that kept its toe and bellies above it, or a step beside the hole that stops short of the floor, leaves too much rock at the bottom and too little halfway up.

Where the thin cover is on the side the toe breaks to, the usual fixes pull against each other: bringing the hole forward thins the cover, moving it back adds to the toe. The findings then say so — stem the hole down past the thin level, and handle the toe with a toe hole or a heavier bottom charge. Fix the front row keeps to the same: it does not bring such a hole forward, leans it only as far as the cover does not thin, and says why. See The front row.

How do I keep the holes from blasting the ore under a waste bench?

Under Top of ore, say where the ore is — a surface of its top, or the block model at or above a cut-off — and the stand-off, the waste each hole leaves over it. Every hole is then stopped that far above the ore. Until you give a stand-off, holes that reach the ore are flagged and the checks ask for one. See Waste over ore.

Why does the blast already know its floor and crest?

Because the survey does. Choose a boundary (or type the corners) on a pit with a surveyed surface and the ground under the block is read for the levels it stands at. A bench is taken from one level down to the next below it, so every such pair is offered — RL 1100 → RL 1090 · 10.0 m · 4,210 m² — and the one standing on most of the block is taken for you. Pick another and the floor, the crest and the bench height move to it. Where the block sits wholly on one flat, the level its face drops to is read off the ground just outside it.

Typing in Floor RL or Crest RL, or choosing Levels I type, settles the levels yourself and the page stops offering its own. See The levels a boundary stands at.

Why is the floor offered a little below the level the survey reads?

Because that floor is not flat. Where a floor humps and hollows over a metre or more — RL 1089.2 to RL 1091.6, say — with no one level holding enough of it, the bench is offered down to its lower edge, so nearly every hole on it stays in play. Taken at its mean instead, the part of the floor already at or below the mean would be out of play: holes there design to nothing. Under the bench, Where its floor is taken switches between Its lower edge and Its mean, and the page says beside it how much of the block each leaves out of play. To take neither, type a floor of your own. See The levels a boundary stands at.

Can Trackapt take a thin bench, such as a capping over the ore?

Yes, when you ask it to look for one and the survey shows it. A capping of a metre left over the ore, or an ore bench of a metre that truncates on the basement, shows on the survey as flat platforms a metre apart — and left alone, those read as one floor that is not flat. Turn on Expect small benches in the blast's The block window. Where the ground steps between well-defined platforms, the page asks whether they are separate small benches or one floor, and takes no bench until you say. Answered as separate, the thin bench between them is offered like any other. Ground whose heights differ haphazardly, a ramp, and a graded floor are read as one floor and not asked about. Platforms less than half a metre apart, or too narrow for the survey's sampling, are not found; type those levels. A flat-bottomed sump may be asked about too — answer that it is one floor. The switch is off by default, and a saved blast keeps it. See The levels a boundary stands at.

The blast says these levels cannot be drilled. What is it checking?

The levels you gave against the surveyed ground under the block. Four things are impossible on that ground, whoever typed them: a floor at or above the ground (there is nothing under it to drill), a floor more than one bench below it, a crest below all of the ground (the bench above has not been taken yet, so the holes would start under it), and a crest at or below its own floor — that last one is refused when the blast is saved as well.

Anything else is a warning and may well be what you meant: part of the block already dug out, a crest above every collar, a level the survey does not hold, or a bench height that disagrees with the ground. Where the ground offers a real bench near what you typed, Take RL 1100 down to RL 1090 sits beside the message. A block with no surveyed ground is checked for one thing only — the crest against its own floor, which needs no survey — and Save is held until that is right.

Can I see the blast in 3D?

Yes — 3D on the pattern card. Each hole shows its stemming and charge, standing in the blast's surveyed ground, over the grade and any top of ore. Click a layer in the legend to hide it and see the holes inside, or Drill holes to take the holes out and look at the ground on its own. Each collar sits on the surveyed ground where its hole is drilled.

Pictorial, beside it, shows only what a person on the bench would see: the ground, solid, and a collar where each hole is marked out.

After the blast, what are the red lumps scattered over the ground and the heap?

Predicted flyrock: single rocks thrown clear of the heap. One leaves each front-row hole in the way its rock goes (face burst), and three leave the collar of each hole the muckpile says craters. Each lands at the pattern's flyrock range — the figure the flyrock clearance is worked from — on the ground it comes down on, and is never drawn past it. They are sized from the coarse end of the predicted fragmentation and lie where they landed.

They are not misfires: a hole that never fires keeps its column. Their line in the key, Flyrock (predicted), shows while a shot is played in 3D or Pictorial; click it to take them off the picture. See The muckpile it is predicted to leave.

How do I change the view without leaving full screen?

Use the Plan, 3D and Pictorial buttons inside the picture, beside its legend. They are the same three as on the pattern card, so you can switch without scrolling back up, and a picture taken full screen stays full screen. Going between 3D and Pictorial also keeps where you turned and zoomed to. On a phone the buttons show their icons only.

If the legend and those buttons cover the holes you want to look at, click the eye at the picture's top-right corner to take them off the picture; click it again to bring them back. The choice is kept as you switch between Plan, 3D and Pictorial. The expand control beside the eye fills the screen. See The pattern in plan and in 3D.

Can I see a berm or island in 3D, not only on the plan?

Yes. The eye on a berm, island or strip card opens the pattern as a window with that ground shaded, in whichever view the pattern card is showing. The window has its own Plan, 3D and Pictorial buttons inside the picture, and the ground stays shaded in all three. The hole positions left off it are pale rings on a dark rim, so they stand out on any ground. Switching in the window does not change the pattern card's view. See Berms.

Why does a berm that looks bigger have a smaller width?

Because the width is the berm's alone — the flat ledge, measured across as its typical width along its length — and the window also shades the ground behind its crest: the rest of its ragged ledge and the wall up to the benches above. That ground is violet, the berm itself blue-green. A low berm with a tall wall behind it can look far larger than a higher, wider berm with nothing shaded behind it, and ground behind two berms is shaded once, behind the lower. See Berms.

Why were holes left out behind a berm, off the flat itself?

Because they are on the wall side of the berm's crest. A berm is seldom a smooth ledge — faults, joints and weathering leave its crest ragged — so the flat the page finds is narrower than the berm, and holes just behind it would blast the ledge and cast it into the main blast. The draft leaves out every position on ground joined to the berm and no more than 1 m below it (never more than 0.4 of its height over the crest), up the wall behind. The berm's card counts the ones on the ledge itself — its ragged edge, no higher than the berm and that margin — not those up the batter or on the benches above, which no pattern drills; the checks list them, and its window shades all of that ground as Behind the berm's crest. That ground, and the berm itself, is left out of the rock's volume too, so the powder factor is the charge over the rock it is fired into; the card says how many square metres were left out. If the flat is production ground, choose Drill it on its card: nothing behind it is left out then. See Berms.

A berm on my wall dips in places. Is it read as one berm or several?

Where it sags and comes back up, as one. Each stretch either side of a sag is flat enough to read as a terrace of its own while the dip between them is too steep to read as either, so the page joins them: the berm standing either side, the two stretches within a step of each other, the dip at least half a metre below them, and the ledge's own step in front of the dip, falling away to the bench. One card, one Leave it or Drill it, and no hole drafted in the dip. A ramp cut down through a berm joins the same way, so the road down belongs to the berm and its card counts it.

It is not joined where the berm dips and stays down (two berms, two cards), where two berms stand one above the other, or where the top of the bench stands high between two ends of a ledge — ground with no step under it is the top, and it is drafted. Ground below this blast's crest is never a berm. See Berms.

Why did the draft leave holes off the slope?

Because a rig cannot stand there. Holes are only drafted on the top of the bench: not on the face, not well below the crest, and not on the bench above. The front row is leaned towards the face and brought forward only as far as the rig can still stand on top. See Where a hole can be collared.

Why are some of my holes 20 m deep on a 10 m bench?

Because the ground under the block spans more than one bench and nothing said how tall the pit's benches are. Set the Pit bench height and Bench reference RL once under The block and save them for the pit: ground spanning several benches is then offered bench by bench, each drilled to the floor of the one below, and taking them all in one blast is a choice you make. See The pit's planned benches.

Why is no hole drafted on a narrow terrace on the wall?

The terrace is a berm: the catch ledge left where one bench steps back from the next, which keeps falling rock off the floor below. The page finds one on the surveyed ground — a flat about the pit's berm width (no wider than half as much again), standing at least a fifth of a bench above the blast's own crest (set the crest level: without one, no berm is looked for), with the wall falling at least a fifth of a bench (2 m on a 10 m bench) along one side and rising as much along the other — and drafts no hole on it, even a temporary berm on a body mined along its dip. A card on the pattern says left as a berm, and the eye on the card opens the pattern as a window with that berm's ground shaded, so you can see which flat is meant. Ground at the level being blasted — a narrow top, a haul road, a platform you said is a small bench of its own — is never a berm, however rough. An island or strip running up to a berm is still named on its card: only what stands on the berm's own step is left with it, and the rest is reached. If a flat is not a berm — a platform of this bench — choose Drill it on its card and its holes are drafted; until you say, it is left. Set the berm width for the pit under The block (10 m unless you say), or for one blast under Pattern rules: 0 looks for none, otherwise 2 to 100 m. A saved blast shows its berms when reopened, but nothing on it moves. Take a berm only once the benches above it are taken: start at the top bench and work down, or open a new cut along strike from the outcrop. See Berms.

Why did the draft put lighter holes along the wall?

They carry the buffer charge. Beside a final wall or the toe of a berm, the pattern's full charge would break back into rock that has to stay, so no hole stands within half a burden of the wall (2 m at a 4 m burden), and out to 5.2 m the pattern's own holes stay where they are but carry 60 % of their column, the rest stemmed, leaning away from the wall — charged lighter, not removed. A buffer row is laid only in place of the pattern's holes left out inside the stand-off, and a trim row of small decked holes can stand there too; where the pattern left no hole out, no row is laid and its own holes carry the buffer charge. Where the stand-off left ground no hole reaches, a buffer hole is added — but the row and those holes never carry more explosive than the holes they replace, so there is never more explosive near a wall than the pattern alone would put there. Where no hole can be designed there, the draft warns how many places are left and names the holes nearest them. Ground at or above the crest all the way out past the final wall is a wall: that is what a final wall at a pit limit is, the rock beyond it unmined and level with the bench top or higher. Only ground that falls away more than a fifth of a bench below the crest (at least a metre) within a burden and a half of the wall is a free face, and a windrow that rises then falls is a free face too. Turn the rows on or off and set the charge for the pit under The block, or for one blast under Pattern rules — Wall control. A saved blast shows its walls and warnings when reopened, and nothing on it moves. See Wall control.

The blast reached the pit floor — where does mining go next?

Once a blast is on the pit's floor and only berms are left beside it, a card on the pattern says so: mining goes on from the top, widening the pit bench by bench down, and — where the ground runs along strike — in a new cut opened from the outcrop, which it names. Start at the top bench opens a new blast on the pit at its top bench, unsaved and undrafted. See Where mining goes next.

Why did the draft add leaning holes round a patch inside the blast?

The patch is an island: ground no rig can stand on to drill — a hump, a mound too steep, a hollow above the floor — with holes all round it, which would be left standing as a toe. A strip no rig stands on that runs to the edge of the pattern — a step down to a terrace, holes dropped along it — is one too, where holes stand along both sides of it and the ground below it does not fall to the floor: its rock has nowhere to go. A run at the edge with holes along one side only, such as the back row against the face of the bench above, is the edge of the pattern and is left as it is; and nothing against the final wall is taken, as that rock is for the wall's own holes. The draft reaches an island or strip with holes from the ground round it, leaned into it where they need to be, and a card on the pattern says whether they reach all of it. If the island is meant — a ramp, a bund — choose Leave it and the holes are taken out. If they cannot reach all of it, the card says whether the rest is out of a rig's reach (doze it down or level it) or under ground too thin to cover a charge (fill it or level it); treat the area first and draft again. A shorter stemming typed on one of those holes is lengthened back to what its cover needs, and the page says so. See Islands in the pattern.

Can I hide the holes with warnings?

Yes. On a blast, Hide … with warnings beside the view buttons takes them off the plan, the 3D pictures and the plan sheet. On the Drillholes register, once Validate has found something, Hide … with warnings or errors does the same for the plan window. Nothing is removed; the tables still list them.

Can I print the blast pattern to scale?

Yes — Plan sheet on the pattern card. It opens the same sheet survey plans print on: choose the paper, the orientation and the scale, then print at true size or download it. Every hole is marked and named with its depth and, where it is not vertical, its inclination, over the block and the ground's contours. If a scale leaves hole labels running together, or a hole off the sheet, the sheet says so and names a larger scale. See The plan sheet.

Can I print the drillhole database in plan, to scale?

Yes. Open the plan from the sheet icon on the Drillholes register, then Plan sheet above the drawing. Every hole drawn is marked at its collar and named with its depth and, where it is not vertical, its inclination, on the same sheet survey plans print on. See The database in plan.

Can Trackapt tie the shot up, or do I have to connect it myself?

It drafts the tie-up and you own it. Open the Tie-up tile, say how it runs — echelon, row by row, or a V from the centre — and Tie the shot up lays a surface connector between each pair of holes and a downline in each one, choosing the delays nearest to the hole-to-hole and row-to-row milliseconds in the design. Left to itself it starts at the end of the front row, where there is a free face, so every hole after the first breaks towards ground already moved. Pick a different starting hole, tie each row twice, change the downline, or give any hole its own firing time by hand and the network is worked round it. See Tying the shot up.

Can I change one connector without tying the whole shot up again?

Yes. In the Tie-up window every connector is listed; pick one to change its product and delay, or remove it — removing asks first and names any hole that nothing would then reach. Add a connector takes the hole the signal comes from, the hole it fires, and a connector from the catalogue; the app suggests one and your pick stands. On the firing picture, tap a connector to pick it or tap two holes to add one. Every change is checked at once, and Tie it up again asks whether to keep your changes or start clean. See Editing the tie-up by hand.

Why do the cut-off and relief checks only appear once the shot is tied up?

Because until then there is nothing to check. A hole delay and a row delay are a rule about the pattern, not a statement about any hole: on a shock tube shot the signal starts at one collar and runs from there, so a hole's time is the path the signal takes to reach it plus its own downline. Whether a connector passes its signal before the ground beside it moves, whether a hole has anything broken in front of it, and what the delays' scatter does to the charge in one 8 ms window are all questions about that path. See What the checks make of the tie-up.

Can I watch the blast go off before I fire it?

Yes. Watch it fire on the Tie-up window, or turn on the Firing switch beside Plan, 3D and Pictorial once the shot is tied up. On the plan the signal runs along each connector, each hole flashes as it detonates and keeps the colour its firing time gives it, and an arrow shows which way its rock leaves; underneath, the kilograms gone and the kilograms inside the last 8 ms. In 3D and Pictorial the holes fire as tied, the rock flies, craters open and the predicted heap forms, so the muckpile comes on by itself. Play it Very slow, Slow or in Real time — about two seconds — drag the scrubber to any moment, or press Go to the end of the shot for the settled heap. The record button saves the shot as a video. See Watching it fire.

My supplier's connectors are not the ones listed. Can I add my own?

Yes, and you should. The lightning bolt on the Explosives window adds the shock-tube delays in common use — a starter, connectors from 9 to 176 ms, downlines from 400 to 500 ms — as a starting point, marked as figures in common use across the industry rather than your supplier's. Change one and the note goes: from then on it is yours. Anything your supplier makes that is not on that list is added like any other product: name it, set its kind to Detonator, say what it does in the tie-up, and type its delay and scatter off the data sheet. See The delays a shot is tied with.

How far is the muck really going to be thrown?

Further than the app can tell you, or nearer, until you have surveyed some. The predicted muckpile throws each hole's rock the way its timing sends it and lands it, swollen and slumped, on the surveyed ground — and the line under the figures says whether the constants behind it are yours or typical. Until a mine has surveyed its own heaps they are typical, and the shape and the way it changes when you move the start or retie the shot are what is worth reading; the distances are not measurements. See The muckpile it is predicted to leave.

Does the predicted muckpile show a misfire, or a toe that won't break?

Yes. Before firing, a tied-up shot is run the way the tie-up would run it, and any hole behind a connector that moving ground cuts is shown never firing: its rock is drawn standing, in its own colour, and the card names the holes and the connector — C4–C7 never fire (the connector from C3 to C4 is cut by the ground B4 moves): 1,240 m³ left standing. After firing, name the misfired holes on Record the firing and the heap leaves their rock standing instead; any you counted but did not name are said on the card rather than guessed. A front hole's toe with more rock in front of it than its bottom charge can break is left as a hump at the floor, and a hole with no open ground to break to when it fires heaves instead of throwing. The rock constant behind the toe judgement is under Ground and rock, 0.4 kg/m³ unless you set your own. See The muckpile it is predicted to leave.

We fired one hole in solid ground. Why does the heap look like a crater, not a pile?

Because that is what it would be. A charge with no face to break to cannot throw rock forward: it breaks out along the line of least resistance, which is up through the stemming, and leaves a hollow. The muckpile draws that hollow below the ground the hole was drilled from, says how wide and how deep it is predicted to be, and colours the collar for it. Rock the charge breaks but cannot lift out of the hole heaves instead — it swells where it stands and lifts the ground, and is never drawn as thrown. See What each hole's rock does.

I charged 0.5 m of a 10 m hole. Does it still leave a muckpile?

No. Each hole is judged on how far its charge can actually break — Langefors' maximum burden for what that hole holds, scaled down for a charge shorter than six hole diameters. A charge that cannot reach the surface breaks nothing at the surface: nothing is thrown, what it does break heaves where it stands, and the bench above it is left standing and drawn in its own colour. The card says which of those happened, hole by hole.

Can a hole next to an open face say "Craters through the collar"?

No. A short stemming at an open face can still throw fragments out of the collar, but that is a flyrock question, and the flyrock check already answers it by comparing the collar's range against the clearance you need. The muckpile says craters only where there is no face within the charge's reach — a real crater a surveyor could measure. One question, one check.

What do we survey to make the rock's own constants ours?

Two things, both on Reconcile. The muck survey — the loose volume, the solid it came from, the furthest throw and the height — gives the swell and how fast rock leaves the face. The crater survey gives the rock itself: the depth, the diameter, the volume where you measured it, and how far the ground rose around it. A crater is the plainest measurement a blast gives of how far a charge breaks, so it is what the rock constant is fitted from, and the rise of the ground gives how much the rock bulks when it is broken with nowhere to go. Both are fitted per rock domain, because that is a property of the rock, not of the mine — a weathered cap and the fresh rock beneath it are two different constants. A crater test on a single hole counts, and so does a survey that found nothing broke through: file it as 0 and 0. See Reconciling.

We surveyed one crater. What does it tell us on its own?

What the rock constant would have to be for that hollow to be the one your charge left. Fill the crater survey in and the dialog reads it straight back: the constant it implies, and how far the charge actually reached against the reach the design was drawn at. A crater bigger than the one predicted is rock that breaks more easily than the design took it to be, so it reads a lower constant; a smaller crater reads a higher one.

It is a reading, not a calibration, and the line says so where it stands. One crater is one collar, in one piece of ground, under whatever stemming actually went in — a joint through the collar or a wet hole moves it as much as the rock does. Put it beside the next one rather than acting on it: your rock's own constant is worked from five or more surveyed blasts in the same rock, under From your own blasts on that rock domain, where it arrives with the spread it was fitted over. Saving the survey changes no design and no rock domain. Where the reading cannot be made it says which — no crater was predicted for that hole, the hole is not in the design, the survey found no crater, or a figure in it is one the page would not send. See Reconciling.

Where do the fitted figures actually appear, and who accepts them?

On the rock domain, under From your own blasts. Open a domain under Rock domains and the dialog reads the blasts fired in that rock and shows what they fit — the rock constant from the craters, the confined bulking from the heave, the swell from the muck surveys — each with the range the blasts gave, how far they wander, and how many blasts are behind it. Accept into this draft fills a figure in and marks it Calibrated; you still read it and still press Save. Under five surveyed blasts nothing can be accepted, and every blast the fit could not use is named with its reason. See From your own blasts.

We don't survey our muckpiles. Does that break anything?

No. Every muck field on the reconcile dialog is optional, most mines fill in none of them, and the blast reconciles without them. The only difference is that the predicted muckpile keeps saying it is not calibrated.

We do survey them. When do the figures become ours?

After five surveyed blasts, and not before. One muckpile is one reading of ground that varies enormously from blast to blast, so a single survey is shown beside what was predicted and applied to nothing. Once enough blasts agree, the constants are proposed with the spread they were fitted over, with any odd shot named and set aside rather than averaged away, and with the blasts, benches and period they came from written beside them — so you can see what you are being asked to trust. See Reconciling.

Maintenance

A job was completed by mistake. Can it be taken back?

Yes. Open the job and choose Reopen job; you are asked why, and the reason stays in the job's history. It goes back to the status it had just before it was completed, and the next service is measured from the completion before it — unless a later service of the same schedule has been completed since, which still sets the countdown. Nothing is deleted. See Reopening a completed job.

How do I put parts on a job?

Raise your store's issue form from the job itself: on the job, under Stores issues, choose Raise stores issue. Saving it links it to the job, and it is counted there once the storeman files it. Only forms marked Stores issue are offered — whoever builds your forms ticks that on the one your store uses. See Issuing parts to a job.

Is the Stores issues figure a quantity, or a cost?

Neither. It is a count of filed issue documents — one issue can be a split pin or a gearbox, and stores forms carry no price. Drafts are not counted.

Why does Stores issues say "not linked", or "partly or not linked"?

Some jobs have no stores issues linked to them, because their parts went out on paper. not linked means nothing is recorded for them at all, which does not mean no parts were used. partly or not linked beside a number means some of the jobs are like that, so the number is a floor. A plain number is whole. See Where the work went.

Retail

How do I charge VAT at the till?

Under Retail → Setup → Tax codes, add a code — for example VAT, Value added tax, Standard, 15%, with Shelf prices set to Include the tax (the other choice is Tax added at the till). The first code you add is the default, so every product is taxed under it straight away; give zero-rated or exempt products their own code on the product. The till then shows the tax under the total, and the server checks every sale against the tax setup as it stood when the sale was rung up. A rate change only affects sales from its date onward, and the till picks it up within a minute. If the till's tax ever differs from the setup, the line is booked one of two ways: with the till's tax, where the difference is one the setup explains (a code the shop had when the sale was rung up, at a rate and treatment that code carried in the week up to the sale, applied to the price Trackapt booked for the line), or otherwise with the setup's figure. Either way the line is counted in the Tax summary's note and listed on Price audit, which shows what the till charged and what was booked. See Tax codes.

Where do I get the figures for my VAT return?

Retail → Stock reports → Tax summary, set to the return's period: taxable net and tax by code, plus zero-rated, exempt and not-taxed sales, with refunds taken off. Export it to Excel or PDF. See Tax summary.

Can I only take stock by scanning a barcode?

No. A stock count can be reached three ways.

Scan what you are counting takes whatever a handheld scanner types, and on a phone the camera button beside it reads the label instead. Or find it by name takes part of the item's name — or part of its barcode or SKU — and offers the matches to pick from. The quick keys below both boxes are tapped straight through. All three open the same Counting <item name> panel, and you type what is on the shelf.

The last two are there for the things a shop sells that carry no printed code: loose produce, bakery, anything sold by the tray, and any item whose label has been torn off. Such items need no barcode in Trackapt at all — a barcode is optional — and may carry a short SKU of the shop's own instead.

Two things worth knowing:

• Delisted items are offered, marked delisted, in the search list and on the keys. An item you have stopped selling can still be on the shelf, and counting it to zero is how it stops counting as stock.
• Nothing shows you what Trackapt expects. A count that opens on the expected figure is a confirmation rather than a count, and it agrees every time. The gap comes back the instant you record.

See Stock and counting.

Can a cashier find an item by name at the till?

Yes. Beside the scan box is Or find it by name: type part of the name, the barcode or the SKU, and the matches appear with their shelf prices. Click one — or press Enter where exactly one matches — and it goes onto the basket as a scan would. The caret returns to the scan box straight afterwards, so the next scan lands where it should.

There are three ways onto a basket, and they are for different problems:

Reach for	When
The scanner	Normally. It is what the lane is built around.
A quick key	The item carries no printed code at all — loose produce, bakery, a carrier bag.
Find it by name	The label will not read, on something too uncommon to have a key.

Searching needs the connection, exactly as scanning does — both ask Trackapt what the code or the term means. Only the quick keys keep working offline, because the whole grid is fetched when the lane opens so every tap is answered from the device. If the network has dropped mid-shift, the grid is what to reach for.

Delisted items appear in the results, marked delisted. The till refuses to sell one and says so; to take one back, use Return.

Are the stock count's quick keys the same as the till's?

Yes — the same grid, read from the same place. An item is on it because somebody laid it out under Retail → Setup, or, where nobody has, because it is one of the things that location sells most. The line beside the heading says which of the two you are looking at.

A key behaves differently on the two screens, though:

• On the till a key sells the item, a delisted key is refused, and holding it asks for a quantity.
• On the stock count a key takes the item in hand so you can type what is on the shelf. A delisted key works, because a delisted item can still be on the shelf. There is no hold gesture, because the figure is typed into the panel the key opens. And the key shows the item's barcode and SKU (where it has them) rather than the price, since a count has no use for a price.

How do I choose which quick keys the till shows?

Under Retail → Setup → Quick keys. Choose Grid for — every lane, or one location — add products with Add a product, give each key a shorter label or a colour if you like, put them in order with the arrows, and click Save quick keys. A grid holds up to 48 keys, and saving one needs level 3. Until a grid is saved, the till shows Best sellers: the products sold most in the last 60 days. See Laying out the quick keys.

How do I count a whole aisle rather than one item at a time?

Use Retail → Count sheet. Choose a location and a group, click Build the sheet, and every item in that group is listed with an empty box beside it. Type what is on the shelf, leave the lines you have not reached blank, and record.

This is the only way to find an item that is not on the shelf at all. Scanning can only ever report on what somebody thought to scan, so a line that has been empty for a fortnight gets scanned by nobody and counted by nobody, and its stock figure goes on saying whatever the ledger last believed. On a sheet an uncounted line is an empty box rather than an absence.

Three things worth knowing:

• A blank is not a zero. Only the lines you typed a figure into are recorded, so leaving half the aisle for tomorrow costs nothing and records nothing false.
• Searching the sheet never loses a figure. A hidden line keeps what you typed and is still recorded; the screen says how many are hidden.
• The expected figures are still not shown — same reason as everywhere else in counting. The gaps come back the moment you record.

See Count a whole aisle.

Why is a count sheet one group at a time?

Because a whole supermarket on one screen is not a list anybody walks, and because that is how counting is actually sustained — a rolling count of one section a week beats a full shutdown twice a year and finds the problems while they are still small.

There is a practical limit too. A sheet lists up to 2,000 products; a larger group is listed to its first 2,000 in name order, and the screen says so: count the rest on Stock count, one at a time. Record sends every line you counted, in parts of up to 500 lines, all stamped with the same moment, so every line is measured against the ledger as at the moment you first pressed it. If a part is refused, the lines already recorded leave the sheet and the rest stay on it; pressing Record again sends those, still as at your first press, and the gaps shown are the whole sheet's. If the connection drops part-way, a part may have been recorded even though the screen did not hear back — recording it again would count those lines twice, so check with a manager before pressing Record again. See One aisle at a time.

The item I want to count is not in the search. Why?

It has not been made sellable. Counting works over the retail catalogue, and an item joins that catalogue when somebody gives it a price under Retail → Catalogue → Make a product sellable (a barcode is optional). Until then Trackapt has no price to value a gap at, so there is nothing a count could report.

You will also only ever see items in the Tracker Groups you are whitelisted on.

Does every product need a barcode?

No. A barcode is optional. Loose produce, bakery, carrier bags and anything else without a printed code can be made sellable with just a price. Sell them from a quick key, by name, or by a SKU — the shop's own short code, such as BRD-WHT. A product can have a barcode, a SKU, both or neither. A barcode or SKU that is given must be unique in the shop.

What is a SKU, and how does the till use it?

A SKU is your own short code for a product, set in the catalogue: letters, digits, -, _ and ., up to 32 characters, unique in the shop, capitals ignored. Type it into the till's scan box and press Enter — Trackapt tries the code as a barcode first and then as a SKU — or into the name search, where the exact SKU is listed first and Enter rings it up. The catalogue, valuation report and their exports show it in its own column. See Typing a SKU.

Can I set up the whole catalogue at once instead of one product at a time?

Yes. Retail → Catalogue → Import from spreadsheet: download a template for a category, fill in barcodes, SKUs, prices, reorder levels and the rest, and upload it. You see what each row would do before anything is saved, and only the rows without errors are applied. A blank cell leaves a value as it is; a single - clears it. See Setting up many products from a spreadsheet.

How do I find out when a product is running low?

Give it a reorder level in the catalogue. When stock across all your locations gets down to it, the product shows Low stock, appears on Catalogue → Reorder list with its reorder quantity as the suggested order, and your level 3+ team gets one chat alert — not one per sale; it can alert again after a restock. See Reorder levels.

How do I sell something whose price changes every time?

Mark it Open price in the catalogue. The till then opens a price pad whenever it is rung up — scanned, searched, by SKU or from a quick key, online or off — and the cashier types what this one costs. Cancel adds nothing. The line is marked open price, its price can be changed by clicking it, and the sale records the typed price and who rang it up. Hardware checkout lanes cannot ask for a price, so they refuse open-price products. See Open-price products.

How do I give a customer a refund?

Click Return on the till and find the sale — by the receipt number, or by the days they came in and what they bought. Choose how many of each item are coming back (never more than were sold and not already refunded), whether each goes back on the shelf or is written off, and why. Below level 3 a supervisor approves it with their PIN: a team member of this app at level 3 or above who has set a PIN, or the account holder once they have set one. The refund is the price the customer paid. See Refunds.

Can I refund without a receipt?

Only if a level-6 user has set a limit under Retail → Setup → Refunds, and then only at today's shelf price, up to that limit, with a supervisor's PIN. A product sold at an open price cannot be refunded this way: find its receipt. See Refunds without a receipt.

Why does the till ask before removing a line or voiding a basket?

So nothing comes off a basket by accident — and because every void and removed line is recorded against the cashier, with the lane and what came off. Managers read them on Retail → Voids & refunds. See Voids and removed lines.

Departments

Does a Global Metrics group respect departments?

It depends on whether you file it. A Global Metrics group holds no items of its own; its formulas draw on parameters in other groups, which can sit in any department. Left in no department, it is read as cross-departmental and shows under every department lens, which suits a figure such as total operations spending drawn from every department. Filed under one or more departments from its settings, it behaves like any other group and shows only while one of those departments is in focus, which suits a group built solely from one department's parameters. See Departments.

A user is not in a department. Can they see that department's Global Metrics group?

Yes, if the group itself lets them in. Departments organise what is on screen; they never decide who may see a group. A user sees a Tracker Group, global or not, when it is public, when they are on its member list, or when they are the account holder. Being in or out of the department the group is filed under changes nothing. Inside the group, each formula is a second gate: a member sees a figure only when they can see every group it draws from, so a reader may open a Global Metrics group and find some figures withheld. See What a department is not.

Forms

How long can a submitted form be corrected?

Seven days from when it was submitted, unless the form sets its own window. In the form builder, Days a submission can be corrected takes any whole number from 1 to 366. The window applies to deleting as well as correcting, and to submissions already filed as well as new ones — lengthening it reopens older submissions that fall inside the new length. Approving a submission closes it at once, whatever the window. See The edit window.

Who can change how long submissions stay correctable?

Only an administrator — Level 6 or Admin clearance — or the account holder. Other authors of the form see the setting read-only, with a padlock explaining why, and can still change everything else about the form. Every change is recorded under the field in the form builder: the old and new number of days, who changed it and when. See A form can keep its submissions open longer.

Head office only acts on requisitions at month end. By then they are locked. What do we do?

Ask an administrator to give the requisition form a longer window — 45 days covers a month end with room to spare. They open the form in the builder and set Days a submission can be corrected. Because the setting applies to submissions already filed, the requisitions already waiting reopen as soon as it is saved. Two things it cannot change: an approved requisition is closed for good, so head office must vary the quantities before it approves; and once goods have been delivered against a requisition, or it has been closed, it is locked whatever the window says. See A form can keep its submissions open longer.

Why can't I correct a submission that is still inside its window?

Because it has been closed, or because another document has been raised against it — a Goods Delivered answering a Requisition, for instance. Either one locks a submission for everyone, the account holder included, and the submission's screen says which. Reopen it, or withdraw the documents raised against it, before changing it. See Closed and raised-against submissions are locked.

Why can't I move a reading to another date? It says the value came from a form submission.

Because the figure was published by a form entry, and its date and period are that entry's — one figure of a document cannot cover a different day from the rest of it. On the item's readings you can still correct the value and the comment, for as long as the submission itself can be corrected, and the correction is written back into the submission. To move the figure, open the form entry it came from and correct the date there: every figure the entry published moves with it. When the change is turned away nothing is saved, the value included, so make the two changes separately. A figure that was broadcast by hand can still be moved from the item's readings. See Correcting a figure a form published.

Readings

My broadcasts "are safely saved but sync failed due to: Nothing was broadcast: reading 3 of 12 cannot be taken because…". What happened?

The request was malformed at the reading it names, so none of it was recorded — and the message says as much, because the next thought is otherwise "so the others landed". The end of the first sentence says what is wrong: a value that is missing or is not a number, a group, item or parameter that is not a valid id, a missing period or date, a time that is not HH:mm, a note that is not text, or a date or capture time that is in the future (check the device's clock) or is not a real date. The count starts at 1, in the order the readings were sent. Trackapt's own broadcast screens send readings in the right shape, so this is mostly the answer an integration gets. Nothing is guessed — "50" sent as text is not read as 50 — and nothing is deleted from a device's offline queue by a refusal. See When a broadcast is turned away.

Why was the whole broadcast refused when only one reading was wrong?

Because a request that is malformed in one place cannot be trusted in the others, and recording eleven readings of twelve leaves somebody to work out later which one is missing. Either all of it arrived or none of it did. This is different from a reading refused because a form entry already holds that period: there the rest of the broadcast is recorded, and the answer names the form to edit instead. See When a broadcast is turned away.

My comment was cut off and now ends in "…". Why?

It was longer than a comment can be. A comment holds up to 500 characters, and so does a parameter's text value; a longer one is stored cut to fit, ending in … where it was cut so that a reader can see it is not the whole remark. The same length applies wherever the remark came from — a broadcast, a correction, or a form field the form broadcasts — because they all land in the same place and are read side by side. The reading itself is always recorded — Trackapt does not turn a figure away because the remark beside it ran long. See A long comment is shortened, never refused.

Why does a figure on Period Reports stop before the end of the period I chose?

It is capped at the last date every input has figures. When one of the parameters it is worked out from, a survey for instance, was last recorded before the end of the period, the figure is worked out only up to that date, so days with figures are never set against days without. Its Basis column reads Capped at and the date, and a note under the table names the parameter that set it. Untick Cap at the last date every input has figures in the parameter's settings to work it out over the whole period. See Capped figures.

I changed a target part-way through the month. What happens to the readings already filed?

Each reading is judged by the target in force on its own day, everywhere it is shown. With 100 t a shift from the 1st and 120 t from the 15th, the shifts of the 1st to the 14th are held to 100 and the rest to 120, so the month's plan is the two added up. The dashboard, Reports → Period, Supervise and the parameter's page all show the target as typed, 100 → 120, with a line saying Target changed on 15 Oct: 100 → 120. To start the change on another day, pick it under Takes effect from when you change the target. See Changing the plan.

Can I set a target that starts next week, or one that should have started last week?

Yes. When you change the target on the tracker group, pick the day under Takes effect from. A later day leaves today's target alone until that day, and the field shows Target changes to 120 on 20 Oct meanwhile. An earlier day re-judges the readings from that day on, and the screen says which ones before you save. Every change is listed under Target history on the parameter's page. See Changing the plan.

I typed the wrong target and corrected it the same day. Which one counts?

The correction. Two changes for the same day count in the order they were made, so the later one is the target, and both stay listed under Target history.

Why can't I type the unit of a computed parameter?

Its unit is worked out from its formula, so it always matches what the formula measures. To use a different one, pick it from the unit picker under Result, or add it to the unit library there. See The unit and the plan of a computed parameter.

Time and attendance

Why can I not see a worker somebody else added?

Usually because you are not named as one of their supervisors. In this module that list is not a note about who somebody reports to — it is the rule that decides who the record exists for. You see your own attendance and the attendance of the people you are named a supervisor for, and nothing else.

Clearance does not help by itself: a site manager who is not named on a worker does not see that worker's hours, and is not meant to. To see them, be added to the list. On Attendance → Workers, press Supervisors on that worker's row and add the name. Doing that needs level-5 clearance, so if the button is greyed, hover it — it says so — and ask somebody who holds it, or the account holder.

If you are named and still cannot see them, you cannot see the Employees group their record is in — both are needed. See Say who may see each worker.

I added a worker. Do I have to name myself as their supervisor?

No. Whoever puts a person on the register — with Enrol from group or Add person — is made their supervisor automatically, so the ordinary case, a foreman enrolling the casual who started on Monday, needs no action at all. The Supervisors field in that dialog is for naming other people as well as yourself, and it is offered only at level-5 clearance. See Put your people on the register.

Where do I type a worker's name and employee number?

On their record in your Employees group — not in Attendance. Each person on the register is a record in a group with Attendance · Workers ticked under Used by, and their name, employee number and site are read from it. Change them there once and every attendance screen follows. The worker editor shows them read-only, with Edit in and the group's name to take you to the record.

For somebody with no record yet, Add person on Workers creates the record in the group and enrols them in one go. It needs level-4 clearance, the level that adds records to a group. See Label your Employees group.

Why are the names in capitals, and will that upset payroll?

Names read in capitals everywhere in Trackapt, the attendance screens and every export included — the payroll export too, for every row. It changes nothing in payroll: hours are matched to people by employee number, never by name.

A row says "not mapped" in the Payroll check column. What do I do?

Nobody has confirmed which property of that person's group holds the employee number, so payroll has nothing to match the row on. On Workers, open Employee fields and press Confirm on that group. Trackapt suggests the property from its name and marks it guess; nothing is used until it is confirmed. Confirming needs level-5 clearance, because it changes what payroll prints. See Confirm the Employee fields.

Why will a person, or their whole group, not delete?

Somebody in it is on the attendance register and has clock records. Deleting the record would orphan hours that payroll may still need, so Trackapt refuses and names who is in the way. Mark them as left in Attendance instead — their hours are kept, and they drop off the live screens.

The same applies to taking Attendance · Workers off a group's Used by: it is refused while somebody in the group has clock records and belongs to no other group used for Attendance workers. Move them into another Employees group first; after that the change goes through.

Why is there no Delete on some items?

Those items hold readings. An item with readings can be deleted only by the account holder or an admin, so other members are offered Delete only on items that hold none.

When the account holder or an admin deletes an item with readings, a dialog says how many readings go with it and asks for the item's name to be typed. Delete stays unavailable until the name matches. Deleted readings cannot be recovered.

An item with no readings asks for a plain confirmation: Delete or Keep.

A worker's Supervisors column says "nobody — only the account holder sees them". Is that a fault?

It means no supervisor can see that worker, while they carry on clocking in every morning. Their hours are being recorded correctly; nobody except the account holder can look at them.

A notice at the top of Workers counts them, so it is not something you have to go looking for. Press Set supervisors on each row and name somebody. The screen refuses to save an empty list. See Workers.

The Supervisors button on a row is greyed out. Why?

Because changing who may see a worker takes level-5 clearance — the same bar as naming somebody to a department, and higher than the level 3 that keeps the register. It is not an ordinary edit: a name added to that list gains a standing view of another person's attendance.

Hover the button and it says which of two cases you are in: below the bar, or below the bar and not a supervisor of that worker either. Both send you to the same place — somebody with level-5 clearance, or the account holder. You can still read the whole register, enrol workers and add shift patterns.

If the register itself looks short, that is the access rule rather than a missing control: you are seeing the workers you supervise plus your own record, and the screen says so above the list.

What happens if I take myself off a worker's supervisors?

That worker leaves your register, and you will not be able to put yourself back — restoring a name requires being able to see the worker, so only a colleague who still can will be able to do it.

Trackapt asks before it happens and names who will stop seeing the worker, including you where that applies. Changing the list in the row does nothing on its own; Save is the press that moves access. See Say who may see each worker.

Micro-finance

Can every loan have a different interest rate and repayment schedule?

Yes — that is how the module is built. There is no price list: the rate, whether interest is flat or on the reducing balance, how often the borrower repays (daily, weekly, fortnightly, monthly or every so many days), any days nobody collects, and any fee are agreed with that borrower and stored on that loan. See Agreeing the terms.

What is the difference between flat and reducing-balance interest?

Flat charges interest on the whole amount lent for the whole term, and spreads it evenly over the instalments. Reducing balance charges interest only on what is still owed each period, so the interest shrinks as the loan is repaid. At the same quoted rate, flat costs the borrower much more — which is why every quote shows the True yearly cost beside it.

What is the "true yearly cost"?

What the loan really costs a year once the fee, the interest method and the timing of every payment are all counted — the effective annual rate. "5% a month, flat" repaid over four months comes to roughly 140% a year. Tell the borrower this figure. See The true yearly cost.

How is a business assessed?

On the five Cs of credit — character, capacity, capital, collateral and conditions — the framework every lender uses, weighted the way micro-lenders weight it: character and capacity count for most. The answers give a score out of 100 and a grade from A (low risk) to E (decline). See Assessing a business.

What does "flight risk" mean?

The risk that the borrower disappears before the loan is repaid. It is read from how rooted the owner and business are: how long they have traded from this spot, whether the premises are fixed, where they live, whether family live nearby, and whether they plan to move. A high flight risk never grades better than D — a referral to the credit committee — however good the books look.

What is DSCR?

The debt-service coverage ratio: the business's free cash each month divided by its monthly repayment. 1.25 or more is the usual comfort level; under 1 the business cannot repay from its own cash, and the loan is declined. See Can they afford it?

Can a business assess itself?

Yes. On the borrower's page, Send a self-assessment link and send it to the owner by WhatsApp or SMS. They answer on their own phone with no login; the link works once, for 14 days. A loan officer then verifies the answers on a visit — a self-assessment is never enough on its own to approve a loan. See The self-assessment.

Why can't I approve a loan I raised?

Because the person who negotiated a loan should not be the one who signs it off — separating the two is what any lender's auditor looks for first. Another approver (level 5) must approve it. The same goes for a loan you did not raise but renegotiated or chose the collector for while it was an application: changing it made it yours. A lender run by the account holder alone may approve one they raised or changed, and the loan records that it did.

If the application is changed by anyone after you opened it, the approval is refused and asks you to reload and review — so nobody approves terms they never saw.

How do collectors know who to collect from?

Micro-finance → Collections lists the overdue borrowers first, longest late at the top, then today, tomorrow and each day up to the collector's own Look ahead, with the amount, the phone number and where to find each one. With Morning message on, the same list arrives as a notification each morning. See Collecting.

Can a collector choose how far ahead they see what is due?

Yes. Look ahead on the Collections screen is each collector's own setting — from today only up to a month ahead — and their morning message covers the same days.

What is PAR30?

Portfolio at risk over 30 days: the whole outstanding principal of every loan more than 30 days late, as a share of all the money out on loan. It is the industry's headline health figure — under 5% is sound, over 10% is trouble. See The portfolio.

A payment was recorded against the wrong loan. Can it be deleted?

Not deleted — reversed. An approver opens the loan and presses Reverse beside the payment, with a reason. It stays on the loan, marked reversed, and stops counting; then record it again against the right loan.

Can we charge borrowers for paying late?

Yes, if it was agreed. Late charges are set on the application with the other terms: either a fixed charge per late instalment, or a percentage of what is late for each 30 days it stays late, with grace days before the first charge and a cap on the total. They are made each morning, paid only after the arrears themselves, and never charged on an earlier charge. See Late charges.

Can a late charge be let off?

Yes. An approver (level 5) opens the loan and presses Waive beside the charge, with a reason — for all of what is left, or part of it. The charge stays on the loan with who waived how much and why.

A borrower has fallen behind. Do we have to write the loan off?

No — it can be restructured. A loan officer proposes new terms for what is still owed and says why; another approver approves it. The unpaid instalments are replaced by a new schedule on the same loan, the unpaid interest is carried over (never added to the principal), and the loan is marked restructured. See Restructuring a loan.

Why is the Portfolio showing a "Restructured" figure?

Because a restructure starts a loan's days late again from its new schedule, so without its own line a book could look healthier just by rescheduling everything that falls behind. Supervisors watch the restructured share for exactly that.

Does the module send borrowers reminders?

Yes: a reminder before each instalment, an overdue notice once a week while a payment is late, and a receipt for every payment. Only by notification on a phone where the borrower turned reminders on, or by email at an address they confirmed — never SMS, and never to an address that was only typed in. See Reminders and receipts for borrowers.

Why hasn't a borrower received an email?

Their address has to be confirmed first. On the borrower's page, Send verification email; nothing is emailed until they open the link in it. If the address on file was changed, it has to be confirmed again. The borrower's page also lists every message sent and how it went out.

How does a borrower turn reminders on?

Press Reminders link on the borrower's page and send them the link from your own phone. On their phone they open it and press Turn on reminders on this phone. They can turn them off, or stop all messages, from the same page.

Does it keep accounts for the lending?

Yes — Ledger (approvers, level 5) is a double-entry ledger of the lending: payouts, repayments split into principal, interest, fees and late charges, recoveries, reversals and write-offs, with a trial balance and a journal you can export. It is kept on a cash basis. See The ledger.

Can a ledger entry be corrected?

Not edited — entries are never changed or deleted. Reverse the payment that was wrong; the reversal posts its own entry, and the right payment posts another.

Chat

Can colleagues see when I am online, or when I was last on?

Only people you already have a conversation with. Under the name at the top of a one-to-one chat they see online while you have Trackapt open, and last seen today at 9:05AM (or yesterday, or a date) once you close it. Someone in your organisation you have never messaged sees nothing, and nobody outside your organisation ever does.

How do I stop people seeing my last seen?

Open the chat, press ⋮ at the top and choose Hide my last seen. From then on nobody sees when you are online or were last on. It works both ways, as in WhatsApp: while yours is hidden you will not see anyone else's either. Choose Show my last seen from the same menu to turn it back on.

How do I know someone is replying?

While they type, the top of the chat says typing… (in a group, who is typing). It also shows in the conversation list, in place of the last message.

How do I copy a message?

Right-click the message (on a phone, press and hold it) and press the Copy button. "Message copied" confirms it. A link in a message is copied with its address, so pasting it somewhere else still leads to the page.

How do I reply to a message?

On a phone, swipe the message to the right until the reply arrow appears, then let go (as in WhatsApp). On a computer, right-click the message and press Reply. The message you are answering shows above the message box; press Esc or the ✕ to cancel.

How do I react to a message?

Right-click the message (on a phone, press and hold it), press the smiley and pick one of 👍 ❤️ 😂 😮 😢 🙏, or press + for any other emoji. Your reaction shows on the bottom edge of the message for everyone in the chat, in one small bubble with up to three emoji and how many people reacted. Tap that bubble to see who reacted with what. You have one reaction per message: pick another to change it, or tap Tap to remove under your name to take it back. Reactions on a deleted message go with it.

How do I add emoji to a message?

Press the smiley at the left of the message box. Pick as many emoji as you like: each goes where the cursor is, so you can put them anywhere in the message. Search by name ("thumbs", "heart", "thanks"), or open Recent for the ones you use most. Press the smiley again to close it.

Can I make a voice or video call?

Yes, one to one. Open a chat with the person and press the phone (voice) or camera (video) button at the top. The first time, your browser asks to use the microphone (and camera): allow it. Their screen rings wherever they have Trackapt open; they press Answer or Decline. During the call you can mute, turn your camera off, switch between front and back camera on a phone, and End call. The chat keeps a line for every call — "Missed voice call", "Video call · 3:05" — with a button to call again.

A few things to know:

• With chat notifications turned on, a call also rings as a notification — Incoming voice call (or video call) — even when Trackapt is closed. Tap it while it rings to open Trackapt and answer. If nobody answers, it changes to Missed voice call; tap it to open the chat and call back. If you answer or decline on another device, it quietly changes to Answered or Declined, and it disappears from any device where Trackapt is open.
• Your screen says Calling… until the other person's phone or computer has the call, then Ringing… — as on WhatsApp. If none of their devices can be reached (switched off, no connection, Trackapt not installed for notifications), you see Not available after a few seconds and they get a missed call. Someone already on a call shows On another call. This means a caller can tell whether you are reachable, even if you hide your last seen.
• Calls are not available in groups or from a shared desk — call the person directly.
• When the devices can reach each other directly, the call goes straight between them. Otherwise it is relayed, encrypted, through Cloudflare's relay service; the relay cannot listen in, but it does see both devices' network addresses and when the call took place.
• On every call, relayed or not, each side's network address (including an address on a local network) is visible to the other device once the call is answered, as with any direct call.
• If you keep calling someone who does not answer, Trackapt stops placing calls to them for a few minutes. Send a message instead.
• When the two devices are on different networks (mobile data, another office), the call is relayed through a secure relay service so it still connects. A relayed call can run for at least 12 hours, a full long shift. If a call says Couldn't connect, check your connection, try Wi-Fi, or send a message instead.

Can I make the chat text bigger?

Yes, the chat has its own text size. Open the chat, press ⋮ at the top and, under Text size, pick one of the four A buttons, from Small to Extra large. The messages and the message box change at once, so you can compare before closing the menu. Nothing else in Trackapt changes size. This is on top of Trackapt's own text size: with the whole app set large, the chat can still be made larger again. The choice is kept on this device only.

What if I send a message with no signal?

It waits on your phone with a clock (⏳) beside it and goes out by itself as soon as you are back online. Nothing is lost if you close Trackapt in the meantime. If it cannot be sent, it shows a red ⊖: tap it to try again. A message that did reach the other person while your connection was dropping is recognised the next time you open the chat, and the red mark goes away on its own. The other person never receives it twice.

Does Enter send the message?

On a computer, yes — press Shift+Enter for a new line. On a phone or tablet Enter starts a new line and you press the send button. Esc cancels a reply or an edit, and ↑ in an empty message box edits the last message you sent, while it can still be edited.

I started typing in one chat and opened another. Did I lose it?

No. What you had written stays with that chat as a Draft, shown in red in the conversation list, and is back in the message box when you return. Drafts are kept on this device until you send them or close Trackapt.